Here's an idea to save a few (fractions of?) a cent at every transaction: don't actually log anything, rely on deterministic evaluation of transactions to do all the logging virtually.
This however, requires quite expensive and error-prone evaluation of each and every transaction on the chain, to see whether any of its subtransactions affect the contract we care about. Not possible until we have something like an advanced yet robust version of VulcanizeDB or equivalent on which to run such queries.
Also, logs might still be useful to make some the past occurrence of some events somewhat cheaper to verify from another contract, though if we know which they are, we might also put them in the contract state and make it query-able (though in case of high contention on the contract, this technique can make verification unreliable and/or expensive).
File under optimization techniques that will remain out-of-reach for a few years, anyway.
Note that when porting to Bitcoin Cash and possibly other platforms, we won't have actual logging, only virtual logging, so the feature will be welcome.
Here's an idea to save a few (fractions of?) a cent at every transaction: don't actually log anything, rely on deterministic evaluation of transactions to do all the logging virtually.
This however, requires quite expensive and error-prone evaluation of each and every transaction on the chain, to see whether any of its subtransactions affect the contract we care about. Not possible until we have something like an advanced yet robust version of VulcanizeDB or equivalent on which to run such queries.
Also, logs might still be useful to make some the past occurrence of some events somewhat cheaper to verify from another contract, though if we know which they are, we might also put them in the contract state and make it query-able (though in case of high contention on the contract, this technique can make verification unreliable and/or expensive).
File under optimization techniques that will remain out-of-reach for a few years, anyway.